It’s a simple, utopian idea. If we give everyone a monthly check, we can eliminate poverty and do away with the inefficiencies of our cumbersome and flawed welfare state. Minneapolis is the latest city to give a “universal basic income” (UBI) a try. It’s offering $500 a month for 18 months to 150 of its low-income residents with no work or spending restrictions.

But others worry it’s not so simple. A universal basic income would be expensive, and what if it discourages people to work, which could inadvertently increase inequality and lead to social instability? A new paper suggests the skeptics may be right: UBI may cause more harm than good for a very high cost.

Testing UBI is not easy. What makes UBI universal and basic is that everyone gets the money, and the flow of cash is predictable and long-lasting. UBI advocates point to a few experiments that show giving people checks doesn’t cause them to work less.

But the Minneapolis experiment and other studies aren’t truly UBI because they’re short term; the payments only last a year or two. And that impermanence fundamentally changes how people will respond. Most financial and work decisions are based on the outlook for lifetime income, not a few dozen months of extra cash.

How the payments are structured is also important. Another widely cited study looks at income paid each year from the Alaska Permanent Fund. The economists estimate the payments haven’t caused Alaskans to decrease work, and may even encourage beneficiaries to do more part-time work. But the permanent fund isn’t true UBI either because payments are based on the state’s oil revenues and thus vary significantly year to year. So the permanent-fund payments actually increases Alaskans’ income risk, which is the opposite of what UBI is supposed to do.

Advocates might say these studies are close enough. But close is not enough in this case. Claiming the models represent UBI is like saying a fixed-rate consol bond is a close approximation to a two-year bond or a dividend-paying stock. We know from finance that these types of assets have completely different values and can therefore elicit completely different behavior on the part of investors in terms of risk taking and wealth effects.

A new study from the National Bureau of Economic Research takes a different approach to evaluating UBI. The economists reviewed lottery winners over a five-year period. Lottery winners are a good test for UBI because lottery winnings are large enough that the income they generate can be life-changing. The economists estimate the average winnings are equivalent to an extra $7,800 a year, similar to UBI proposals. Lottery winners are also chosen at random, which makes for a good experiment.

Contrary to unlocking creativity, motivation and entrepreneurship, the economists estimate lottery winners are unlikely to start a successful business. They also estimate that winners worked less and were more likely to change jobs to one paying a lower wage. The economists also observed many winners moved soon after the lottery, usually to a more rural area. But few moved to a higher-quality neighborhood, in terms of college attainment of neighbors, average income and other metrics that are a proxy for opportunities available to them or their children. There was one positive effect: lottery winners are more likely to marry and less prone to divorce.

You may think living in the country and working less isn’t so bad. Working a lower-paid job can sometimes offer other benefits, like flexibility and time with your children. But there are costs. Working less at a less-demanding job often means you forgo learning new skills and wage increases. This may not be a big deal for people in middle age. But it can leave young people who are still establishing their careers and acquiring skills much worse off. Most wage increases occur in your 20s and 30s, and if you miss out on those years, odds are you won’t catch up.

Our current welfare system is imperfect. But the fact that it makes payments contingent on earnings, age or even having a child is a better alternative. First of all, it’s much cheaper because you don’t have to give money to the many people who don’t need it. Second, guaranteed money (or money you get no matter what happens) is worth much more than income that only pays off some of the time. The more money we give people, the bigger impact it will have on their behavior, and often not in a good way.

One issue with the recent lottery study is that it only tracks winners for several years. Many lottery winners think the windfall will set them up for life, but they end up filing for bankruptcy and are then prone to depression and bad health (which is why you should always take the annuity option). But this demonstrates the challenges of implementing UBI. Once it’s offered, it’s very difficult to take it away, and doing so can leave people worse off than they started.